• Reminder: There is no character clause. We do not care if they might have gambled on the game or taken PEDs. Nor do we care if they devoted time and money to charity. Again, this is only about performance on the field.
• Once we had everyone's scores, we tallied them up and ranked players based on their average score.
Wins above replacement (WAR) is a valuable tool when making an argument, but when arguing which players are the greatest of all time, it only measures career value. Peak value, as coined by Bill James, is a little different, measuring the height of a player's greatness rather than his breadth. Take for example, the case of Sandy Koufax, who ranks 74th in career pitching WAR at FanGraphs. Even the most hard-core stathead alive won't try to claim that Koufax wasn't greater than Frank Tanana, Chuck Finley and Jerry Koosman, all of whom have a higher career WAR.
That's where GAR comes in. What GAR does is put career WAR in a historical context that takes into consideration both a player's career value and peak value. It starts with career WAR and adds a player's five-year peak WAR, multiplied by 1.6 to put peak and career on an equal scale. For a baseline, replacement level doesn't make sense -- typical replacement level is talent that's freely available, which just won't do when trying to separate the great from the greatest. Instead, we've chosen as the baseline the average of the 20th through 30th best at each position, that sweet spot at which you've stopped talking about inner-circle Hall of Famers and started talking about the fictional Hall of Very Good.
GAR is a blunt instrument, and it is not intended to end arguments but simply to add an additional statistical tool to look at when we argue about whether X was a greater ballplayer than Y, which we do endlessly this time of year. -- Dan Szymborski